The demo looked simple.
Two black VDI modules on the table. A compact Keysight VNA in the center. A gold waveguide straight-through calibration standard between them. A laptop showing an S21 curve.
And then, to the right: a red and black mechanical arm holding a VDI module at its end, pointing it at different angles under software control.
The measurement hardware was straightforward.
The mechanical arm was the interesting part.
What was on the bench
Keysight P5006B portable VNA — the compact field-deployable vector network analyzer, visible with its distinctive red lightning bolt logo — serving as the measurement backbone. The P5006B covers up to 53 GHz natively, requiring frequency extenders to reach W-band.
Two VDI WR10+ modules (67–115 GHz) — connected to the P5006B via blue coaxial cables, with a gold WR10 waveguide calibration standard between them. These are the frequency extenders that push the VNA's measurement range into the W-band, enabling full S-parameter characterization across 67–115 GHz.
One VDI WR8.0 module (90–140 GHz) — the flat, low-profile unit visible on the table, extending coverage further into D-band.
The official VDI EuMW 2025 placard specifications: → W-Band Mini VNA Extender (TxRx) in smaller housing → High Test Port Power → Compatible with MilliBox Antenna Test Systems and Positioners → Additional frequency bands available upon request
MilliBox GIM04 300E — the red and black multi-axis robotic positioner, with "MilliBox" labeled on the central rotary joint. The GIM04 is a 4-axis gimbal-style positioner, 3D-printed in red and black polymer composite, with precision servo motors at each joint. At the end of the arm: a VDI module with a gold horn antenna or waveguide aperture pointing toward a reference antenna or calibration target.
The laptop showed a live measurement: an S21 curve with a characteristic Gaussian-shaped peak — the antenna gain pattern being swept as the MilliBox arm rotated the DUT through angular positions. The S21 curve peak corresponds to the boresight gain. The rolloff on both sides traces the antenna's radiation pattern in the measurement plane.
A second laptop in Image 1 showed the radiation pattern as a 2D heat map — the orange and red peak surrounded by a dark background, the spatial distribution of W-band energy visible as a color gradient.
What the MilliBox GIM04 actually measures
The GIM04 is not a standard lab turntable.
Standard turntables rotate a DUT around a single axis — they can sweep azimuth angle, one plane at a time. To measure a full 3D radiation pattern, you need multiple single-axis sweeps, repositioning the setup between measurements.
The GIM04 is a multi-axis gimbal positioner. With 4 degrees of freedom, it can orient the DUT in any direction within its workspace without repositioning the bench setup. The arm reaches, rotates, tilts, and positions the mounted VDI module to any angle defined by the software control sequence.
At W-band and above, this capability is essential for several reasons:
The antenna patterns are narrow. A W-band antenna with 20 dBi gain has a half-power beamwidth of approximately 10–15°. A single-axis turntable with coarse angular steps will miss the main lobe entirely if the step size exceeds the beamwidth. The GIM04 allows fine angular resolution — the "300E" designation likely refers to 300-step-per-revolution encoder resolution, giving sub-degree positioning accuracy.
The DUT is small. At 90–115 GHz, the antenna under test may be a few millimeters across — a small horn, an on-package patch array, or a waveguide aperture on an IC evaluation board. Mounting these on a precision arm and sweeping them programmatically is the only practical way to map their radiation characteristics.
The measurement environment matters. At W-band, reflections from the test bench surface, cable movement, and nearby objects all appear in the radiation pattern measurement. The MilliBox arm keeps the measurement geometry consistent across the sweep, minimizing systematic errors from environmental interaction.
The compact form factor: what it changes
Traditional W-band antenna measurement systems — anechoic chambers with large turntables, far-field ranges, or near-field scanners — are expensive, space-intensive, and operationally complex.
The VDI + MilliBox combination on this bench fits on a 1.5-meter table.
This is not incidental. It represents a specific market position: the researcher or engineer who needs W-band antenna characterization capability but cannot justify a dedicated chamber. A university lab developing 6G chip-scale antenna arrays. An automotive radar team that needs to characterize prototype modules. A satellite communication group testing waveguide transitions.
The "compact" in "Compact W-Band VNA Extender" is not a size description. It is a design philosophy — bringing W-band measurement capability out of the dedicated test facility and onto the lab bench.
The price implications are significant. A full anechoic chamber with W-band range capability costs €500,000 to several million euros. The VDI + MilliBox GIM04 system is orders of magnitude less expensive, at the cost of measurement accuracy in highly absorptive or low-reflectivity environments where the open-bench setup will pick up environmental contributions.
For prototyping and early-stage characterization, the tradeoff is favorable.
Why W-band antenna measurement is geometrically hard
At lower frequencies — say, 10 GHz — the wavelength is 30 mm. A 1 mm positioning error in the antenna-to-antenna separation represents 0.033 wavelengths of path length error. This contributes a negligible phase error to the S21 measurement.
At 90 GHz — the lower end of the WR8.0 band — the wavelength is 3.33 mm. The same 1 mm positioning error now represents 0.3 wavelengths. That is a 108° phase error. In an S-parameter magnitude measurement, this would cause measurement uncertainty. In a coherent antenna pattern measurement, it would corrupt the phase response entirely.
This is why the MilliBox arm's positioning accuracy matters — and why the arm must be mechanically stiff, thermally stable, and free of backlash. A 3D-printed structure raises questions about all three of these, compared to machined aluminum. The MilliBox design addresses this by using over-constrained joint structures and high-resolution encoders, at the cost of higher part count and assembly complexity.
The gold waveguide components visible at the DUT end of the arm are machined to tight dimensional tolerances — the WR10 waveguide flange flatness specification is typically ±0.005 mm. At 90 GHz, a 0.01 mm gap in the flange mating surface introduces measurable return loss. The mechanical arm must maintain this flange contact throughout its range of motion.
The two-system configuration
Looking at the bench carefully, there were actually two VNA extender configurations:
Configuration 1: Two WR10+ modules (67–115 GHz) connected to the P5006B with a gold WR10 calibration standard — this is the standard two-port S-parameter measurement setup, calibrated at the waveguide reference plane.
Configuration 2: The MilliBox arm holding a VDI module — this is the antenna pattern measurement setup, where one module is fixed (the reference antenna or source) and the other is swept through angles on the arm (the DUT).
The two configurations demonstrate the flexibility of the VDI extender architecture: the same frequency extension hardware can be reconfigured from a bench S-parameter measurement setup to a spatial characterization setup by moving modules between fixed positions and the positioner arm.
What the radiation pattern display was showing
The laptop in Image 1 showed a 2D color map — an orange-red peak surrounded by a dark blue background. This is the received S21 power plotted as a function of the two angular axes swept by the MilliBox arm.
The concentrated orange-red peak represents the main beam of the antenna. Its shape and width define the antenna's directivity in that angular plane. The dark blue surroundings represent the sidelobe region — where the antenna radiates little power.
The S21 curve on the second laptop showed the same data as a 1D slice: the gain profile along a single angular cut through the main beam.
At W-band, the information visible in these two displays tells the engineer: → Where is the beam pointing? (The peak position) → How wide is the beam? (The 3 dB beamwidth) → How strong are the sidelobes? (The ratio of peak to first sidelobe level) → Is the pattern symmetric? (The shape of the main lobe) → Are there unexpected secondary lobes from structural reflections?
Each of these has direct implications for the antenna's performance in its target application — whether that is a 6G base station sector antenna, an automotive radar front end, or a millimeter-wave imaging aperture.
The engineering shift this represents
Articles 049 and 050 in this EuMW series described systems focused on signal quality and device characterization — OTA link EVM, D-band load pull, PA compression behavior.
This demo added a dimension that neither of those addressed: spatial behavior.
At W-band and above, the system is not confined to the circuit. It extends into the physical space around it. The antenna's radiation pattern is part of the system response — not an add-on, not a post-processing step, but a primary characterization requirement as fundamental as S-parameter measurement.
The MilliBox arm on the demo table was not showing off. It was acknowledging a reality of W-band engineering that lower-frequency designers rarely encounter:
The system is partly inside the hardware, and partly in the geometry.
Mapping that geometry — characterizing how the system behaves across all the spatial configurations it will encounter in deployment — requires tools that were once confined to specialized facilities.
The VDI + MilliBox combination brings that capability to the lab bench.
That accessibility changes who can characterize W-band antenna systems, and when in the development cycle they can do it.
All photos: Thomas · @SignalByThomas